“The vendor-tracing module assumed more log access than my firm actually had, so I had to improvise with screenshots. The sampling worksheets still transferred cleanly into our Q3 KYC application review.”
Niall P.
Notebooks
What people kept after the lab
These notes name modules on this yard. They are not a ratings widget. One of them is deliberately unimpressed about log access.
★★★★☆
Control Ledger as a spine for second-line paper. Kept the event-to-fee drill. Lost a star because the evening sessions collided with our month-end.
Priya marked my population memo as “theatre.” She was right: I had picked photogenic fails. Module 02 forced a proper skip log. I still think module 05 spends too long on tone for people who already write for a committee.
“I sat Watchlist Screening on the waitlist description and then took the flagship instead. The join between a screening hit and a frozen balance was the hole in our last file.”
Case note: e-money onboarding fees
A three-person internal audit team from a UK e-money issuer arrived with a finding that “KYC completeness was adequate.” After the lab they re-opened the population around “verification complete” events that never generated the contracted onboarding fee. The unmatched set was 41 items in a fortnight — small in isolation, enough to rewrite the control as completeness of fee income, not completeness of selfies.
They did not implement a new vendor. They changed the sample instruction and the assertion. That is the shape of work we consider a successful seat.
Case note: documentary lane vs funded status
A solo reviewer at a challenger-shaped retail book used the Documentary Evidence Bench waitlist description, then worked module 04 of the flagship lab on her own time. Her firm’s application showed green ticks on address documents while 12 of her 25 funded accounts had no matching “docs accepted” event in the export. The case study she sent us (sanitised) became a teaching example of cut-off: the ledger believed the customer was onboarded a day before the KYC application did.
Limitation, which she asked us to print: she still cannot extract vendor-side hashes, so the finding stays at application-export level. We agree that is the honest ceiling.